NyanGuard AI Gradio Demo

NyanGuard AI is a local-first moderation demo for AITuber / VTuber live chat and audio streams. It accepts an audio or video file, transcribes speech locally when a local ASR backend is available, replaces toxic language with Nyan Nyan, and returns a playable moderated audio file.

Setup

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -r requirements.txt

Run

python app.py

Then open the local Gradio URL printed in the terminal.

Local AI Notes

• faster-whisper is used for local speech recognition when installed. The first ASR run may download the selected Whisper model.
• pyttsx3 is used for local text-to-speech when available.
• If ASR is not available, paste a transcript in the "Optional transcript" field to keep the demo fully local and deterministic.
• The app is designed so the model layer can be replaced by an AMD Ryzen AI / ONNX Runtime / LFM backend without changing the UI.